HELP!

I have multiple PC's, multiple ATI GPU's, multiple Windows-only OS's.

The previous same-name- ... minus "revisited" ... BOINC thread ended nowhere (last post dated: 17 Nov 2010:- e.g., "could".)

So - why do BOINC Project ATI GPU tasks continue to run when the user is active despite preferences that are set to suspend / "Use GPU while computer is in use" is unclicked ... then the current GPU task runs until it ends ... upon which sanity is then happily restored?
I'm happy to ignore replies like "switch to XYZ OS / GPU's" TIA.

I vote for restoring GPU use to the user immediately when required for non-BOINC work based on one simple rule :- If I want to use my PC I don't want to be interruped by BOINC.
Period.
If that means work ... ummm do it, pretty please???

Ray
(Apologies if I've got any of this wrong - I just click buttons until something works then when it doesn't ... I ask for help.)

---

The difference between 0 and 1 is greater than the difference between 1 and 1,000,000,000

Also make sure that boinctray.exe is running. This does the idle detection for BOINC under Windows Vista and 7.

Showing that it works....
24/11/2011 02:39:00 | | ATI GPU 0: ATI Radeon HD 4700/4800 (RV740/RV770) (CAL version 1.4.1417, 1024MB, 992MB available, 2000 GFLOPS peak)
24/11/2011 02:39:00 | | OpenCL: ATI GPU 0: ATI RV770 (driver version CAL 1.4.1417, device version OpenCL 1.0 AMD-APP-SDK-v2.5 (775.2), 1024MB)
24/11/2011 02:39:00 | | ATI GPU is OpenCL-capable
24/11/2011 06:58:01 | Albert@Home | [coproc] ATI instance 0: confirming for p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0
24/11/2011 06:58:54 | Albert@Home | [checkpoint] result p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0 checkpointed
24/11/2011 06:59:01 | Albert@Home | [coproc] ATI instance 0: confirming for p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0
24/11/2011 06:59:06 | LHC@home 1.0 | [checkpoint] result w5_weak8_collision_err_bb__20__s__64.31_59.32__12_14__6__49.5_1_sixvf_boinc27668_2 checkpointed
24/11/2011 07:00:00 | | Suspending computation - time of day
24/11/2011 07:00:00 | Albert@Home | [cpu_sched] Preempting p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0 (removed from memory)
24/11/2011 07:00:00 | Albert@Home | [task] task_state=QUIT_PENDING for p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0 from request_exit()
24/11/2011 07:00:01 | Albert@Home | [task] Process for p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0 exited
24/11/2011 07:00:01 | Albert@Home | [task] task_state=UNINITIALIZED for p2030.20100913.G44.55+00.20.S.b4s0g0.00000_1672_0 from handle_premature_exit

---

I researched the boinctray suggestion - see thread from 2010 ... http://boinc.berkeley.edu/dev/forum_thread.php?id=5464
which ended "I'll pass" and also covered preferences.

In the meantime having had the chance to casually observe this sticky WU behaviour for a while I have concluded that, for my setups at least, it's a per-project thing as follows:- http://boinc.berkeley.edu/dev/forum_thread.php?id=7135&nowrap=true#41706
Advice there, please.

---

The difference between 0 and 1 is greater than the difference between 1 and 1,000,000,000